Jaume Solà i Pujols - Departament de Filologia Catalana ...

languages. If so, in 0 en Joan would not be adjoined to CP, but rather to IP.
There are some ways out of this problem if we want to maintain the Wh-to-COMP
hypothesis for 0. One could be the following:
a) Interrogative CPs, unlike relative or [-WH] CPs, allow adjunction.
b) The fact that the adjuncts cannot follow the Wh-element in 0 could be due to V+INFL
movement to C o . So the prediction would be that IP adjuncts should follow the verb.
As for the assumption in a), I do not know of any interesting way to derive it. In addition
this assumption does not receive support from other languages: English, French or German
interrogative CPs do not allow adjunction (at least when embedded). As for b), we will see below
that a V-to-COMP account of the adjacency requirement between interrogative Wh-phrases and
the verb is problematic in Romance languages.
Another possibility would be to exploit CP-recursion. We could assume, with Chomsky
(1986-a), that adjunction to Arguments is forbidden. If the embedded CP is an Argument of the
main verb, then adjunction to it is forbidden. Suppose, however, that:
IP ] ].
a) there is CP-recursion: a CP can optionally subcategorize for another CP: [ CP C o [ CP
b) when there is CP recursion, the 'that' particle and the relative Wh-phrases have to
occupy the upper CP, while interrogative Wh-phrases have to (or may) occupy the lower CP.
c) the lower CP is not, strictly speaking, an Argument of the main verb, but rather a
subconstituent of it, so adjunction to the lower CP is allowed.
The assumptions in a) and b) would hold only of Romance NSLs, for the asymmetry
between 0/0 and 0 is not found in English, French or German. So, in addition to the unclear
nature of these assumptions, we should ask why they are only relevant for some languages. I
languages: free adjunction to CP in main clauses would break the
V-2 appearance of the language, contrary to fact. This
restriction is not likely to be universal, however (recall what
we said about Occitan). It could rather follow from parametric
options affecting the nature of CPs. Once a language forbids
this adjunction, however, it seems to be a strong prohibition.
1